Reinforcement Learning describes the situation of a machine learning system, where the only training signal provided by the environment is that of success or failure of the agent, after the system has acted over a sequence of decision cycles. This learning problem can be formulated as a Markov Decision Process (MDP) within the framework of Dynamic Programming. The main motivation behind the Brainstormers' effort in the soccer domain is to investigate Reinforcement Learning (RL) methods in complex domains and to develop new variants and practical algprithmus. We consider it important that we not only demonstrate the principal feasibility of RL, but actually do apply learned behavior in our competition team. Our long term goal is a team of learning agent, where we only plug in 'Win the match' - and our agents learn to generate the appropriate behavior.

• Tutorial on Reinforcement Learning for Soccer Agents
  Have a look at this tutorial (requires Shockwave Flash) to get to know something on the basics of Reinforcement Learning and its application in the Robotic Soccer context. The tutorial illustrates how a soccer-playing agent learns to acquire some fundamental behaviors as well as cooperative behavior using Reinforcement Learning.
  
• Learning to Be Competitive
  

You can find our publications that are related to the Brainstormers 2D project on a separate page.

You may also be interested in the downloads we offer.